I am trying to understand, what should i do in order to load a custom model made with the images i want to use, but i can’t find a clear answer what in the world should i do… First of all, i was looking for a software that i can simply put the images, and then get the model, and load it into opencv, but i couldn’t find literally anything…
What should i do then? Should i download something like tensorflow and make the training software?
what exactly do you need ? classification ? object detection ? segmentation ?
please be more precise
maybe have a look at google’s colab, where all of this is already installed, and where you can use free GPUs for training
I want to detect a around 40 objects, and that’s it, it would be a pain to use cascades inside opencv, that’s why i would need something else
the problem with these things is that i am not a python guy, i do everything in c/c++, and that’s the only language i know. If the thing that you posted is rather copy & paste, and run, then i might try it, do you know which software/freamwork i need to download, in order to run that code?
darknet is still in C
(and no, you dont want to build tensorflow on your box, or use it’s c++ api, believe me ;- )
what kind of data do you have ?
even if you retrain an existing network, you would need some 50 annotated (!) images per class